Artificial intelligence has crossed a significant threshold: it is now contributing directly to real laboratory experimentation rather than merely analyzing data after the fact. In controlled tests reported by Axios, OpenAI’s GPT-5 was used in collaboration with human researchers to improve a standard molecular biology cloning protocol. When its recommendations were tested in an actual wet lab, the process efficiency increased dramatically—by as much as 79 times in some cases.

What makes this development notable is not the speed improvement alone, but the nature of the contribution. The AI proposed procedural changes that human researchers had not initially considered, and those changes held up when tested with real biological materials. This suggests that advanced AI systems may soon become active collaborators in experimental science rather than passive tools.

The implications are far-reaching. If AI systems can reliably assist in designing experiments, optimizing protocols, and identifying overlooked variables, research timelines in medicine, genetics, and biotechnology could compress significantly. At the same time, this raises important questions about oversight, validation, and scientific accountability as AI systems move closer to the core of the discovery process.

Source: Axios

Link:
https://www.axios.com/2025/12/16/openai-gpt-5-wet-lab-biology